神经网络在磨辊离心铸造缺陷诊断中的应用研究

摘要

Building the defect diagnosis model of centrifugal casting grinding roller with the theory of neural network,then the reasonable defect diagnosis result of the centrifugal casting grinding roller was obtained after back-propagation neural network＇s establishing and training.All of the BP neural network＇s error index accord with the demands,which illustrate the reliability and advantage of the models used in centrifugal casting.%应用神经网络的理论和方法,首先建立离心铸造缺陷诊断的神经元模型,然后建立BP神经网络模型进行学习训练后,得出了离心铸造磨辊的正确诊断方案。训练结果表明,该神经网络的各项误差指标均达到了满意的要求,充分有力地说明了用这种BP神经网络模型诊断离心铸件具有可靠性与优越性。